To make it possible to search data stored in a file server, a search server has been conventionally creating an index. The index is used for performing an efficient search of data. To create an index, the search server accesses a file server to read out data attributes, such as names, dates and times of update, and owners, and contents of all data files stored in the file server. The search server reads out the contents of data files for the sake of full-text search. Therefore, index creation by the search server puts a heavy load on the file server.
The index creation has conventionally been performed for creating an index for use in data search using a dedicated search server, and hence it is possible to perform scheduling of the index creation. Accordingly, the search server performs index creation during a time of day, such as late midnight hours, in which no problem is caused even if a heavy load is placed on the file server.
Further, for example, to reduce load placed by the search server, there has been proposed a file server that identifies a search server by a specific protocol or an IP address, and returns only an updated file list in response to an access from the search server (PTL (Patent Literature) 1).